TY - JOUR
TI - Efficient plasma-surface interaction surrogate model for sputtering processes based on autoencoder neural networks
AU - Gergs, Tobias
AU - Borislavov, Borislav
AU - Trieschmann, Jan
T2 - Journal of Vacuum Science & Technology B
AB - Simulations of thin film sputter deposition require the separation of the plasma and material transport in the gas phase from the growth/sputtering processes at the bounding surfaces (e.g., substrate and target). Interface models based on analytic expressions or look-up tables inherently restrict this complex interaction to a bare minimum. A machine learning model has recently been shown to overcome this remedy for Ar ions bombarding a Ti-Al composite target. However, the chosen network structure (i.e., a multilayer perceptron, MLP) provides approximately 4Â106 degrees of freedom, which bears the risk of overfitting the relevant dynamics and complicating the model to an unreliable extent. This work proposes a conceptually more sophisticated but parameterwise simplified regression artificial neural network for an extended scenario, considering a variable instead of a single fixed Ti-Al stoichiometry. A convolutional β-variational autoencoder is trained to reduce the high-dimensional energy-angular distribution of sputtered particles to a low-dimensional latent representation with only two components. In addition to a primary decoder that is trained to reconstruct the input energy-angular distribution, a secondary decoder is employed to reconstruct the mean energy of incident Ar ions as well as the present Ti-Al composition. The mutual latent space is hence conditioned on these quantities. The trained primary decoder of the variational autoencoder network is subsequently transferred to a regression network, for which only the mapping to the particular low-dimensional space has to be learned. While obtaining a competitive performance, the number of degrees of freedom is drastically reduced to 15 111 (0.378% of the MLP) and 486 (0.012% of the MLP) parameters for the primary decoder and the remaining regression network, respectively. The underlying methodology is very general and can easily be extended to more complex physical descriptions (e.g., taking into account dynamical surface properties) with a minimal amount of data required.
DA - 2022/01//
PY - 2022
DO - 10.1116/6.0001485
DP - DOI.org (Crossref)
VL - 40
IS - 1
SP - 012802
J2 - Journal of Vacuum Science & Technology B
LA - en
SN - 2166-2746, 2166-2754
UR - https://avs.scitation.org/doi/10.1116/6.0001485
Y2 - 2021/12/23/14:35:15
ER -